Asymptotically optimal strategies in repeated zero-sum games with incomplete information